libavformat/rtpdec_h264.c

   1 /*
   2  * RTP H264 Protocol (RFC3984)
   3  * Copyright (c) 2006 Ryan Martell
   4  *
   5  * This file is part of Libav.
   6  *
   7  * Libav is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * Libav is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with Libav; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 /**
  23  * @file
  24  * @brief H.264 / RTP Code (RFC3984)
  25  * @author Ryan Martell <rdm4@martellventures.com>
  26  *
  27  * @note Notes:
  28  * Notes:
  29  * This currently supports packetization mode:
  30  * Single Nal Unit Mode (0), or
  31  * Non-Interleaved Mode (1).  It currently does not support
  32  * Interleaved Mode (2). (This requires implementing STAP-B, MTAP16, MTAP24,
  33  *                        FU-B packet types)
  34  */
  35
  36 #include "libavutil/attributes.h"
  37 #include "libavutil/base64.h"
  38 #include "libavutil/avstring.h"
  39 #include "libavcodec/get_bits.h"
  40 #include "avformat.h"
  41
  42 #include "network.h"
  43 #include <assert.h>
  44
  45 #include "rtpdec.h"
  46 #include "rtpdec_formats.h"
  47
  48 struct PayloadContext {
  49     // sdp setup parameters
  50     uint8_t profile_idc;
  51     uint8_t profile_iop;
  52     uint8_t level_idc;
  53     int packetization_mode;
  54 #ifdef DEBUG
  55     int packet_types_received[32];
  56 #endif
  57 };
  58
  59 #ifdef DEBUG
  60 #define COUNT_NAL_TYPE(data, nal) data->packet_types_received[(nal) & 0x1f]++
  61 #else
  62 #define COUNT_NAL_TYPE(data, nal) do { } while (0)
  63 #endif
  64
  65 static const uint8_t start_sequence[] = { 0, 0, 0, 1 };
  66
  67 static void parse_profile_level_id(AVFormatContext *s,
  68                                    PayloadContext *h264_data,
  69                                    char *value)
  70 {
  71     char buffer[3];
  72     // 6 characters=3 bytes, in hex.
  73     uint8_t profile_idc;
  74     uint8_t profile_iop;
  75     uint8_t level_idc;
  76
  77     buffer[0]   = value[0];
  78     buffer[1]   = value[1];
  79     buffer[2]   = '\0';
  80     profile_idc = strtol(buffer, NULL, 16);
  81     buffer[0]   = value[2];
  82     buffer[1]   = value[3];
  83     profile_iop = strtol(buffer, NULL, 16);
  84     buffer[0]   = value[4];
  85     buffer[1]   = value[5];
  86     level_idc   = strtol(buffer, NULL, 16);
  87
  88     av_log(s, AV_LOG_DEBUG,
  89            "RTP Profile IDC: %x Profile IOP: %x Level: %x\n",
  90            profile_idc, profile_iop, level_idc);
  91     h264_data->profile_idc = profile_idc;
  92     h264_data->profile_iop = profile_iop;
  93     h264_data->level_idc   = level_idc;
  94 }
  95
  96 static int sdp_parse_fmtp_config_h264(AVFormatContext *s,
  97                                       AVStream *stream,
  98                                       PayloadContext *h264_data,
  99                                       char *attr, char *value)
 100 {
 101     AVCodecContext *codec = stream->codec;
 102     assert(codec->codec_id == AV_CODEC_ID_H264);
 103     assert(h264_data != NULL);
 104
 105     if (!strcmp(attr, "packetization-mode")) {
 106         av_log(s, AV_LOG_DEBUG, "RTP Packetization Mode: %d\n", atoi(value));
 107         h264_data->packetization_mode = atoi(value);
 108         /*
 109          * Packetization Mode:
 110          * 0 or not present: Single NAL mode (Only nals from 1-23 are allowed)
 111          * 1: Non-interleaved Mode: 1-23, 24 (STAP-A), 28 (FU-A) are allowed.
 112          * 2: Interleaved Mode: 25 (STAP-B), 26 (MTAP16), 27 (MTAP24), 28 (FU-A),
 113          *                      and 29 (FU-B) are allowed.
 114          */
 115         if (h264_data->packetization_mode > 1)
 116             av_log(s, AV_LOG_ERROR,
 117                    "Interleaved RTP mode is not supported yet.\n");
 118     } else if (!strcmp(attr, "profile-level-id")) {
 119         if (strlen(value) == 6)
 120             parse_profile_level_id(s, h264_data, value);
 121     } else if (!strcmp(attr, "sprop-parameter-sets")) {
 122         codec->extradata_size = 0;
 123         av_freep(&codec->extradata);
 124
 125         while (*value) {
 126             char base64packet[1024];
 127             uint8_t decoded_packet[1024];
 128             int packet_size;
 129             char *dst = base64packet;
 130
 131             while (*value && *value != ','
 132                    && (dst - base64packet) < sizeof(base64packet) - 1) {
 133                 *dst++ = *value++;
 134             }
 135             *dst++ = '\0';
 136
 137             if (*value == ',')
 138                 value++;
 139
 140             packet_size = av_base64_decode(decoded_packet, base64packet,
 141                                            sizeof(decoded_packet));
 142             if (packet_size > 0) {
 143                 uint8_t *dest = av_malloc(packet_size + sizeof(start_sequence) +
 144                                           codec->extradata_size +
 145                                           FF_INPUT_BUFFER_PADDING_SIZE);
 146                 if (!dest) {
 147                     av_log(s, AV_LOG_ERROR,
 148                            "Unable to allocate memory for extradata!\n");
 149                     return AVERROR(ENOMEM);
 150                 }
 151                 if (codec->extradata_size) {
 152                     memcpy(dest, codec->extradata, codec->extradata_size);
 153                     av_free(codec->extradata);
 154                 }
 155
 156                 memcpy(dest + codec->extradata_size, start_sequence,
 157                        sizeof(start_sequence));
 158                 memcpy(dest + codec->extradata_size + sizeof(start_sequence),
 159                        decoded_packet, packet_size);
 160                 memset(dest + codec->extradata_size + sizeof(start_sequence) +
 161                        packet_size, 0, FF_INPUT_BUFFER_PADDING_SIZE);
 162
 163                 codec->extradata       = dest;
 164                 codec->extradata_size += sizeof(start_sequence) + packet_size;
 165             }
 166         }
 167         av_log(s, AV_LOG_DEBUG, "Extradata set to %p (size: %d)!\n",
 168                codec->extradata, codec->extradata_size);
 169     }
 170     return 0;
 171 }
 172
 173 // return 0 on packet, no more left, 1 on packet, 1 on partial packet
 174 static int h264_handle_packet(AVFormatContext *ctx, PayloadContext *data,
 175                               AVStream *st, AVPacket *pkt, uint32_t *timestamp,
 176                               const uint8_t *buf, int len, uint16_t seq,
 177                               int flags)
 178 {
 179     uint8_t nal;
 180     uint8_t type;
 181     int result = 0;
 182
 183     if (!len) {
 184         av_log(ctx, AV_LOG_ERROR, "Empty H264 RTP packet\n");
 185         return AVERROR_INVALIDDATA;
 186     }
 187     nal  = buf[0];
 188     type = nal & 0x1f;
 189
 190     assert(data);
 191     assert(buf);
 192
 193     /* Simplify the case (these are all the nal types used internally by
 194      * the h264 codec). */
 195     if (type >= 1 && type <= 23)
 196         type = 1;
 197     switch (type) {
 198     case 0:                    // undefined, but pass them through
 199     case 1:
 200         if ((result = av_new_packet(pkt, len + sizeof(start_sequence))) < 0)
 201             return result;
 202         memcpy(pkt->data, start_sequence, sizeof(start_sequence));
 203         memcpy(pkt->data + sizeof(start_sequence), buf, len);
 204         COUNT_NAL_TYPE(data, nal);
 205         break;
 206
 207     case 24:                   // STAP-A (one packet, multiple nals)
 208         // consume the STAP-A NAL
 209         buf++;
 210         len--;
 211         // first we are going to figure out the total size
 212         {
 213             int pass         = 0;
 214             int total_length = 0;
 215             uint8_t *dst     = NULL;
 216
 217             for (pass = 0; pass < 2; pass++) {
 218                 const uint8_t *src = buf;
 219                 int src_len        = len;
 220
 221                 while (src_len > 2) {
 222                     uint16_t nal_size = AV_RB16(src);
 223
 224                     // consume the length of the aggregate
 225                     src     += 2;
 226                     src_len -= 2;
 227
 228                     if (nal_size <= src_len) {
 229                         if (pass == 0) {
 230                             // counting
 231                             total_length += sizeof(start_sequence) + nal_size;
 232                         } else {
 233                             // copying
 234                             assert(dst);
 235                             memcpy(dst, start_sequence, sizeof(start_sequence));
 236                             dst += sizeof(start_sequence);
 237                             memcpy(dst, src, nal_size);
 238                             COUNT_NAL_TYPE(data, *src);
 239                             dst += nal_size;
 240                         }
 241                     } else {
 242                         av_log(ctx, AV_LOG_ERROR,
 243                                "nal size exceeds length: %d %d\n", nal_size, src_len);
 244                     }
 245
 246                     // eat what we handled
 247                     src     += nal_size;
 248                     src_len -= nal_size;
 249
 250                     if (src_len < 0)
 251                         av_log(ctx, AV_LOG_ERROR,
 252                                "Consumed more bytes than we got! (%d)\n", src_len);
 253                 }
 254
 255                 if (pass == 0) {
 256                     /* now we know the total size of the packet (with the
 257                      * start sequences added) */
 258                     if ((result = av_new_packet(pkt, total_length)) < 0)
 259                         return result;
 260                     dst = pkt->data;
 261                 } else {
 262                     assert(dst - pkt->data == total_length);
 263                 }
 264             }
 265         }
 266         break;
 267
 268     case 25:                   // STAP-B
 269     case 26:                   // MTAP-16
 270     case 27:                   // MTAP-24
 271     case 29:                   // FU-B
 272         av_log(ctx, AV_LOG_ERROR,
 273                "Unhandled type (%d) (See RFC for implementation details\n",
 274                type);
 275         result = AVERROR(ENOSYS);
 276         break;
 277
 278     case 28:                   // FU-A (fragmented nal)
 279         buf++;
 280         len--;                 // skip the fu_indicator
 281         if (len > 1) {
 282             // these are the same as above, we just redo them here for clarity
 283             uint8_t fu_indicator      = nal;
 284             uint8_t fu_header         = *buf;
 285             uint8_t start_bit         = fu_header >> 7;
 286             uint8_t av_unused end_bit = (fu_header & 0x40) >> 6;
 287             uint8_t nal_type          = fu_header & 0x1f;
 288             uint8_t reconstructed_nal;
 289
 290             // Reconstruct this packet's true nal; only the data follows.
 291             /* The original nal forbidden bit and NRI are stored in this
 292              * packet's nal. */
 293             reconstructed_nal  = fu_indicator & 0xe0;
 294             reconstructed_nal |= nal_type;
 295
 296             // skip the fu_header
 297             buf++;
 298             len--;
 299
 300             if (start_bit)
 301                 COUNT_NAL_TYPE(data, nal_type);
 302             if (start_bit) {
 303                 /* copy in the start sequence, and the reconstructed nal */
 304                 if ((result = av_new_packet(pkt, sizeof(start_sequence) + sizeof(nal) + len)) < 0)
 305                     return result;
 306                 memcpy(pkt->data, start_sequence, sizeof(start_sequence));
 307                 pkt->data[sizeof(start_sequence)] = reconstructed_nal;
 308                 memcpy(pkt->data + sizeof(start_sequence) + sizeof(nal), buf, len);
 309             } else {
 310                 if ((result = av_new_packet(pkt, len)) < 0)
 311                     return result;
 312                 memcpy(pkt->data, buf, len);
 313             }
 314         } else {
 315             av_log(ctx, AV_LOG_ERROR, "Too short data for FU-A H264 RTP packet\n");
 316             result = AVERROR_INVALIDDATA;
 317         }
 318         break;
 319
 320     case 30:                   // undefined
 321     case 31:                   // undefined
 322     default:
 323         av_log(ctx, AV_LOG_ERROR, "Undefined type (%d)\n", type);
 324         result = AVERROR_INVALIDDATA;
 325         break;
 326     }
 327
 328     pkt->stream_index = st->index;
 329
 330     return result;
 331 }
 332
 333 static PayloadContext *h264_new_context(void)
 334 {
 335     return av_mallocz(sizeof(PayloadContext) + FF_INPUT_BUFFER_PADDING_SIZE);
 336 }
 337
 338 static void h264_free_context(PayloadContext *data)
 339 {
 340 #ifdef DEBUG
 341     int ii;
 342
 343     for (ii = 0; ii < 32; ii++) {
 344         if (data->packet_types_received[ii])
 345             av_log(NULL, AV_LOG_DEBUG, "Received %d packets of type %d\n",
 346                    data->packet_types_received[ii], ii);
 347     }
 348 #endif
 349
 350     av_free(data);
 351 }
 352
 353 static av_cold int h264_init(AVFormatContext *s, int st_index,
 354                              PayloadContext *data)
 355 {
 356     if (st_index < 0)
 357         return 0;
 358     s->streams[st_index]->need_parsing = AVSTREAM_PARSE_FULL;
 359     return 0;
 360 }
 361
 362 static int parse_h264_sdp_line(AVFormatContext *s, int st_index,
 363                                PayloadContext *h264_data, const char *line)
 364 {
 365     AVStream *stream;
 366     AVCodecContext *codec;
 367     const char *p = line;
 368
 369     if (st_index < 0)
 370         return 0;
 371
 372     stream = s->streams[st_index];
 373     codec  = stream->codec;
 374
 375     if (av_strstart(p, "framesize:", &p)) {
 376         char buf1[50];
 377         char *dst = buf1;
 378
 379         // remove the protocol identifier
 380         while (*p && *p == ' ')
 381             p++;                     // strip spaces.
 382         while (*p && *p != ' ')
 383             p++;                     // eat protocol identifier
 384         while (*p && *p == ' ')
 385             p++;                     // strip trailing spaces.
 386         while (*p && *p != '-' && (dst - buf1) < sizeof(buf1) - 1)
 387             *dst++ = *p++;
 388         *dst = '\0';
 389
 390         // a='framesize:96 320-240'
 391         // set our parameters
 392         codec->width   = atoi(buf1);
 393         codec->height  = atoi(p + 1); // skip the -
 394     } else if (av_strstart(p, "fmtp:", &p)) {
 395         return ff_parse_fmtp(s, stream, h264_data, p, sdp_parse_fmtp_config_h264);
 396     } else if (av_strstart(p, "cliprect:", &p)) {
 397         // could use this if we wanted.
 398     }
 399
 400     return 0;
 401 }
 402
 403 RTPDynamicProtocolHandler ff_h264_dynamic_handler = {
 404     .enc_name         = "H264",
 405     .codec_type       = AVMEDIA_TYPE_VIDEO,
 406     .codec_id         = AV_CODEC_ID_H264,
 407     .init             = h264_init,
 408     .parse_sdp_a_line = parse_h264_sdp_line,
 409     .alloc            = h264_new_context,
 410     .free             = h264_free_context,
 411     .parse_packet     = h264_handle_packet
 412 };